Delete generate_train_test_split.py

generate_train_test_split.py

@@ -1,108 +0,0 @@
-# This file is used to generate the train and test split of files within
-# the file system.
-
-import os
-import re
-import shutil
-import tarfile
-from math import floor
-from random import shuffle
-
-
-def get_file_list_from_dir(data_dir):
-    """Get a list of files within the directory"""
-    all_files = os.listdir(os.path.abspath(data_dir))
-    data_files = list(filter(lambda file: file.endswith('.wav'), all_files))
-    return data_files
-
-
-# Due to the audio and transcript files are in different folders, we need to split
-# them into train and test splits while maintaining their correct mapping in both splits
-# For example, if transcript 0001 is in test split, audio 0001 should also be in test split.
-#
-# Hence, instead of splitting the files itself, I will split the files based on their
-# ID number. Because every transcript and audio file has an ID number.
-# For example, arctic_a0001.txt is the transcript for arctic_a0001.wav.
-#
-# After splitting the ID numbers, we will then assign the files to their respective splits
-# based on their ID numbers.
-def shuffle_file_id(files):
-    file_id_regex = re.compile(r'[a-z]\d\d\d\d')
-    file_id = []
-
-    for file in files:
-        file_id.append(file_id_regex.search(file).group())
-
-    shuffle(file_id)
-    return file_id
-
-
-def train_test_splits(file_ids, training_ratio):
-    split=training_ratio
-    split_index = floor(len(file_ids) * split)
-    train = file_ids[:split_index]
-    test = file_ids[split_index:]
-    return train, test
-
-
-
-if __name__ == "__main__":
-    print("Running")
-    audio_file_paths = {
-        "ABA": "C:\\Users\\user\\OneDrive - Universiti Sains Malaysia\\Assignment\\FYP\\l2arctic_release_v5.0\\ABA\\ABA\\wav",
-        "BWC": "C:\\Users\\user\\OneDrive - Universiti Sains Malaysia\\Assignment\\FYP\\l2arctic_release_v5.0\\BWC\\BWC\\wav"
-    }
-    speaker_accent_map = {
-        "ABA": "ar",
-        "BWC": "zh"
-    }
-    splits = ["train", "test"]
-
-    try:
-        for speaker, accent in speaker_accent_map.items():
-            audio_files = get_file_list_from_dir(audio_file_paths[speaker])
-            shuffled_file_id = shuffle_file_id(audio_files)
-            train, test = train_test_splits(shuffled_file_id, 0.7)
-
-            print(train)
-            print("Number of samples in training: ", len(train))
-            print(test)
-            print("Number of samples in testing: ", len(test))
-            tar_file = tarfile.open(
-                f"C:\\Users\\user\\OneDrive - Universiti Sains Malaysia\\Assignment\\FYP\\data\\audio\\{accent}\\train\\{accent}_train.tar",
-                "w"
-            )
-            for id in train:
-                # Copy training audio into a tar file
-                tar_file.add(
-                    f"C:\\Users\\user\\OneDrive - Universiti Sains Malaysia\\Assignment\\FYP\\l2arctic_release_v5.0\\{speaker}\\{speaker}\\wav\\arctic_{id}.wav",
-                    arcname=f"arctic_{id}.wav"
-                )
-
-                # Copy training transcript into a tsv file
-                with open(f"C:\\Users\\user\\OneDrive - Universiti Sains Malaysia\\Assignment\\FYP\\data\\transcript\\{accent}\\train.tsv", 'a') as tsv_file:
-                    with open(f"C:\\Users\\user\\OneDrive - Universiti Sains Malaysia\\Assignment\\FYP\\l2arctic_release_v5.0\\{speaker}\\{speaker}\\transcript\\arctic_{id}.txt") as txt_file:
-                        tsv_file.write(txt_file.read() + "\n")
-                tsv_file.close()
-            tar_file.close()
-
-            # Copy testing audio into tar file
-            tar_file = tarfile.open(
-                f"C:\\Users\\user\\OneDrive - Universiti Sains Malaysia\\Assignment\\FYP\\data\\audio\\{accent}\\test\\{accent}_test.tar",
-                "w"
-            )
-            for id in test:
-                tar_file.add(
-                    f"C:\\Users\\user\\OneDrive - Universiti Sains Malaysia\\Assignment\\FYP\\l2arctic_release_v5.0\\{speaker}\\{speaker}\\wav\\arctic_{id}.wav",
-                    arcname=f"arctic_{id}.wav"
-                )
-
-                # Copy testing transcript into tsv file
-                with open(f"C:\\Users\\user\\OneDrive - Universiti Sains Malaysia\\Assignment\\FYP\\data\\transcript\\{accent}\\test.tsv", 'a') as tsv_file:
-                    with open(f"C:\\Users\\user\\OneDrive - Universiti Sains Malaysia\\Assignment\\FYP\\l2arctic_release_v5.0\\{speaker}\\{speaker}\\transcript\\arctic_{id}.txt") as txt_file:
-                        tsv_file.write(txt_file.read() + "\n")
-                tsv_file.close()
-            tar_file.close()
-
-    except PermissionError:
-        print("Permission denied")